An EM-Algorithm-Based Approach for Predicting Teacher Candidate Success on the Communication and Literacy Skills Test for Educator Licensure

Abstract

In 1998, the Department of Education in the State of Massachusetts redefined the requirements for teacher licensure, implementing a series of licensure examinations entitled the Massachusetts Tests for Educator Licensure (MTEL). In response, many Massachusetts colleges and universities now offer preparatory support programs to help teacher candidates pass the MTELs. Little research has been conducted to determine the impact, statistical or otherwise, of those measures on students’ MTEL scores. This paper outlines the development and analysis of a linear model for predicting the scores of teacher candidates at a small liberal arts college on the Communication and Literacy Skills Test, which all K-12 teacher candidates in Massachusetts are required to pass. Although failing test scores are reported numerically by the Massachusetts Department of Education, passing test scores are reported only as a “Pass.” In this paper, a variation on the EM algorithm is applied to address the problem of missing data. The statistical technique is outlined in detail and is followed by a discussion of the effectiveness of the preparatory sessions. This paper is accessible to readers who have an introductory level background in statistics.